EPSCoR Research Fellows: NSF: An Explainable AI Supported Performance Monitoring System in Distributed Sustainable Energy Networks

Sustainable energy, such as solar and wind power, due to its commitment to economic, environmental, and social sustainability, has become the preferred energy paradigm for researchers and practitioners in the energy sector, both in the U.S. and globally. However, one of the key concerns in sustainable energy networks is the reliability of performance, which is often affected by unstable power generation and unpredictable anomalies.

To address these issues, the reliability of sustainable energy networks must be improved by understanding new energy characteristics (e.g., solar and wind), designing new frameworks (e.g., decentralized structures), and embracing new technologies (e.g., artificial intelligence (AI)). This project aims to tackle these challenges through a systematic solution: an explainable AI-supported performance monitoring system in distributed sustainable energy networks, which will ensure reliable performance and sustainability.

To achieve this, the PI and a graduate student will visit the host site at the University of Southern California to utilize its cyber-infrastructure, cutting-edge technologies, extensive research experience, and abundant domain expertise. Successful completion of this project will result in the development of an innovative system to improve the reliability of distributed sustainable energy networks, an educational module to advance the teaching and training of UNLV students in the AI/energy field, and enhanced sustainability in energy for the state of Nevada and the U.S.

This project will thereby strengthen the PI’s scientific research, educational capacity, and societal contributions.

The goal of this project is to design a systematic solution for detecting and classifying anomalies in distributed sustainable energy networks using an explainable AI-based method. The project will systematically and experimentally investigate several critical issues in distributed systems, multi-modal learning, and AI explainability. During the investigation, the project will contribute in the following ways: (1) building a hierarchical learning framework capable of processing different local conditions and heterogeneous cluster features for anomaly detection in distributed sustainable energy equipment, (2) proposing a multi-modal learning method that utilizes various factors of sustainable energy, aimed at improving the reliability of anomaly detection and classification with limited labeled data, and (3) developing an explainable AI (XAI) module to reduce the "black box" impact of AI models and facilitate human decision-making in operations.

The project outcomes will advance knowledge and understanding in multi-modal learning and XAI for distributed energy systems and will guide further AI-driven applications to address crucial challenges. The expected results will enrich educational materials and support curriculum development in areas such as distributed systems, multi-modal learning, anomaly detection, and XAI.

Research outcomes will be widely disseminated online, shared at research seminars, and seamlessly integrated with K-12 education and outreach activities, encouraging active participation from underrepresented student groups. This project will enable the PI to develop a long-term collaboration with a nationally prominent institution. Consequently, this project will significantly enhance the PI’s competitiveness as a researcher and educator, improve the research capacity of the PI’s home institution, and contribute to developing a diverse workforce in the PI’s jurisdiction.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.